 Mixing time of a Markov chain

The $\epsilon$-mixing time is the least $t$ for which every initial state has total variation distance at most $\epsilon$ from the invariant distribution after $t$ steps.